In the intersection of Artificial Intelligence (AI) and automated software engineering , there is an abundant potential to reimagine traditional practices and enhance operational efficiency. This paper presents the Six-Tier Architecture for AI-Generated Software Development, an innovative framework built upon the theoretical foundations of automated software engineering. This framework leverages large language models (LLMs) to perform tasks across the software development lifecycle, from high-level abstraction to detailed code generation, thereby systematically improving the automation level of software development. A key aspect of this architecture is the integrated iterative refinement process, which ensures system consistency and enables effective modifications in response to emerging changes. By utilizing the capabilities of LLMs in automated code generation , the Six-Tier Architecture offers a robust, flexible, and dynamic approach to AI-assisted software engineering, illustrating a substantial resilience to the intrinsic changes in software development requirements and processes.